推进技术 ›› 2017, Vol. 38 ›› Issue (5): 1155-1164.

• 控制 测量 故障诊断 • 上一篇    下一篇

基于智能双重响应面法的涡轮叶盘可靠性灵敏度分析

张春宜1,宋鲁凯1,费成巍2,郝广平1,李雪芹3   

  1. 哈尔滨理工大学 机械动力工程学院,黑龙江 哈尔滨 150080,哈尔滨理工大学 机械动力工程学院,黑龙江 哈尔滨 150080,香港理工大学 机械工程系,香港 999077,哈尔滨理工大学 机械动力工程学院,黑龙江 哈尔滨 150080,山东交通学院 工程机械学院,山东 济南 250000
  • 发布日期:2021-08-15
  • 作者简介:张春宜,男,博士,教授,研究领域为可靠性工程及优化。
  • 基金资助:
    国家自然科学基金(51275138);黑龙江省教育厅自然科学基金项目(12531109)。

Intelligent Dual Response Surface Method for Reliability Sensitivity Analysis of Turbine Blisk

  1. School of Mechanical and Power Engineering,Harbin University of Science and Technology,Harbin 150080,China,School of Mechanical and Power Engineering,Harbin University of Science and Technology,Harbin 150080,China,Department of Mechanical Engineering,The Hongkong Polytechnic University,Hongkong 999077,China,School of Mechanical and Power Engineering,Harbin University of Science and Technology,Harbin 150080,China and School of Construction Machinery,Shandong Jiaotong University,Jinan 250000,China
  • Published:2021-08-15

摘要: 为了合理进行整体叶盘多失效模式可靠性分析和准确描述各影响参数的重要程度,将智能算法与双重响应面方法相结合提出可靠性灵敏度分析的智能双重响应面方法(Intelligent Dual Response Surface Method,IDRSM)。首先,建立IDRSM的数学模型,给出基于IDRSM的可靠性灵敏度分析的流程。然后,考虑流场和温度场作用,基于IDRSM对整体叶盘径向变形和应力两种失效模式进行可靠性分析和灵敏度分析。可靠性分析显示:当许用径向变形、许用应力的均值和标准差分别取3.8mm和76μm,690MPa和14MPa时,叶盘综合可靠度为0.9926。灵敏度分析显示:整体叶盘综合失效概率的主要影响因素为流速和转速,占叶盘总失效的92%。通过蒙特卡洛法、响应面法、极值响应面法、智能响应面法等四种方法比较显示:IDRSM能在保证计算精度的前提下提高计算效率。实例分析表明该方法在多失效模式综合可靠性灵敏度分析中的可行性和有效性,也为结构多失效模式可靠性优化开辟了有效途径。

关键词: 可靠性分析;整体叶盘;智能算法;人工神经网络;智能双重响应面法

Abstract: To reasonably implement reliability analysis and describe the significance of influencing parameters on turbine blisk with multi-failure modes,intelligent dual response surface method (IDRSM) was proposed by integrating intelligent algorithm and dual response surface method. Firstly,the mathematical model of IDRSM was established and the basic principle of multi-objective reliability sensitivity analysis with IDRSM was given. And then,based on the proposed method,the reliability analysis and sensitivity analysis of two failure modes (radial deformation and stress) were completed considering the interaction of flow field and temperature field. From the reliability analysis,the comprehensive reliability of turbine blisk is 0.9926 when the mean and standard deviation of the allowable radial deformation and the allowable stress are 3.8mm and 76μm,690MPa and 14MPa,respectively. The sensitivity analysis shows that main impact factors influencing turbine blisk failure are gas velocity and rotational speed,accounting for 92% of the total failure probability. Through the comparison of methods(Monte Carlo method,response surface method,Dual Response Surface Method and IDRSM),it is demonstrated that the proposed IDRSM improves computational efficiency with acceptable computational accuracy. The case study shows the feasibility and effectiveness of this proposed method in multi-failure modes reliability sensitivity analysis,and offers a useful insight into the reliability optimization design of multi-failure mode structure.

Key words: Reliability analysis;Turbine blisk;Intelligent algorithm;Artificial neural network;Intelligent dual response surface method